An Automatic Indexing On-Line Text Retrieval System

摘要

The development and testing of the BROWSER text retrieval system allowing a natural language query statement and providing on-line browsing capabilities through an IBM 2260 display terminal is described. The prototype system contains data bases of 25,000 German language patent abstracts, 9,000 English language patent abstracts, and 8,000 Defense Documentation Center technical abstracts. BROWSER automatically indexes textual documents, creates an inverted file for searching the unformatted text, and creates formatted files for searching bibliographic fields. Bibliographic fields may be searched independently or in conjunction with a text search. In addition to outputs of citations and abstract text, a new form of output - the Response Index - is provided. The response index consists of a one line entry for each abstract retrieved containing the search terms occurring within the abstract. During the browsing phase at the terminal, the response index enables the searcher to view the contents of the file with respect to his search terms and to screen the machine output for the ultimate user. As there are no syntax, or parsing routines the search algorithms are virtually independent of language. The system has been tested on over 100 German language queries. (Author)